def generate_speed_data(traj_dir, grid_idx, feature_path):
MIN_DISTANCE_IN_METER = 5
MAX_DISTANCE_IN_METER = 300
speed_data = np.zeros((grid_idx.row_num, grid_idx.col_num, 1), dtype=np.float)
cnt_data = np.zeros((grid_idx.row_num, grid_idx.col_num, 1), dtype=np.float)
for filename in tqdm(os.listdir(traj_dir)):
if not filename.endswith('.txt'):
continue
traj_list = parse_traj_file(os.path.join(traj_dir, filename))
for traj in traj_list:
for i in range(len(traj.pt_list) - 1):
cur_pt, next_pt = traj.pt_list[i], traj.pt_list[i + 1]
delta_time = (next_pt.time - cur_pt.time).total_seconds()
if MIN_DISTANCE_IN_METER < distance(cur_pt, next_pt) < MAX_DISTANCE_IN_METER:
try:
row_idx, col_idx = grid_idx.get_matrix_idx(cur_pt.lat, cur_pt.lng)
speed = distance(next_pt, cur_pt) / delta_time
# 120 km/h
if speed > 34:
continue
speed_data[row_idx, col_idx, 0] += speed
cnt_data[row_idx, col_idx, 0] += 1
except IndexError:
continue
speed_data = np.divide(speed_data, cnt_data, out=np.zeros_like(speed_data), where=cnt_data != 0)
np.save(os.path.join(feature_path, 'speed.npy'), speed_data)
def generate_transition_view(traj_dir, grid_idx, nbhd_size, nbhd_dist, feature_path):
MIN_DISTANCE_IN_METER = 5
MAX_DISTANCE_IN_METER = 300
meters_per_grid = grid_idx.lat_interval / LAT_PER_METER
radius = int(nbhd_dist / meters_per_grid)
transit_data = np.zeros((grid_idx.row_num, grid_idx.col_num, nbhd_size, nbhd_size, 2),
dtype=np.uint8)
for filename in tqdm(os.listdir(traj_dir)):
if not filename.endswith('.txt'):
continue
traj_list = parse_traj_file(os.path.join(traj_dir, filename))
for traj in traj_list:
for idx in range(len(traj.pt_list) - 1):
cur_pt = traj.pt_list[idx]
next_pt = traj.pt_list[idx + 1]
if MIN_DISTANCE_IN_METER < distance(cur_pt, next_pt) < MAX_DISTANCE_IN_METER:
try:
global_cur_i, global_cur_j = grid_idx.get_matrix_idx(cur_pt.lat, cur_pt.lng)
local_idx = get_local_idx(global_cur_i, global_cur_j, radius, grid_idx, nbhd_dist)
local_next_i, local_next_j = local_idx.get_matrix_idx(next_pt.lat, next_pt.lng)
transit_data[global_cur_i, global_cur_j, local_next_i, local_next_j, 0] = 1
global_next_i, global_next_j = grid_idx.get_matrix_idx(next_pt.lat, next_pt.lng)
local_idx = get_local_idx(global_next_i, global_next_j, radius, grid_idx, nbhd_dist)
local_cur_i, local_cur_j = local_idx.get_matrix_idx(cur_pt.lat, cur_pt.lng)
transit_data[global_next_i, global_next_j, local_cur_i, local_cur_j, 1] = 1
except IndexError:
continue
np.save(os.path.join(feature_path, 'transition.npy'), transit_data)
def generate_line_image(traj_dir, grid_idx, feature_path):
MIN_DISTANCE_IN_METER = 5
MAX_DISTANCE_IN_METER = 300
traj_line_img = np.zeros((grid_idx.row_num, grid_idx.col_num),
dtype=np.uint8)
for filename in tqdm(os.listdir(traj_dir)):
if not filename.endswith('.txt'):
continue
traj_list = parse_traj_file(os.path.join(traj_dir, filename))
for traj in traj_list:
one_traj_line_img = np.zeros((grid_idx.row_num, grid_idx.col_num),
dtype=np.uint8)
for j in range(len(traj.pt_list) - 1):
cur_pt, next_pt = traj.pt_list[j], traj.pt_list[j + 1]
if MIN_DISTANCE_IN_METER < distance(
cur_pt, next_pt) < MAX_DISTANCE_IN_METER:
try:
y1, x1 = grid_idx.get_matrix_idx(
cur_pt.lat, cur_pt.lng)
y2, x2 = grid_idx.get_matrix_idx(
next_pt.lat, next_pt.lng)
cv2.line(one_traj_line_img, (x1, y1), (x2, y2),
16,
1,
lineType=cv2.LINE_AA)
except IndexError:
continue
traj_line_img = cv2.add(traj_line_img, one_traj_line_img)
cv2.imwrite(os.path.join(feature_path, 'line.png'), traj_line_img)
def execute(self, filename, traj_path, mm_result_path):
rn = load_rn_shp(self.rn_path, is_directed=True)
for u, v, data in rn.edges(data=True):
if data['type'] == 'virtual':
data['weight'] = data['length'] * self.alpha
else:
data['weight'] = data['length']
map_matcher = TIHMMMapMatcher(rn, routing_weight='weight')
traj_list = parse_traj_file(os.path.join(traj_path, filename))
all_paths = []
for traj in traj_list:
paths = map_matcher.match_to_path(traj)
all_paths.extend(paths)
store_path_file(all_paths, os.path.join(mm_result_path, filename))
def generate_point_image(traj_dir, grid_idx, feature_path):
pt_cnt = np.zeros((grid_idx.row_num, grid_idx.col_num))
for filename in tqdm(os.listdir(traj_dir)):
if not filename.endswith('.txt'):
continue
trajs = parse_traj_file(os.path.join(traj_dir, filename))
for traj in trajs:
for cur_pt in traj.pt_list:
try:
row_idx, col_idx = grid_idx.get_matrix_idx(cur_pt.lat, cur_pt.lng)
pt_cnt[row_idx, col_idx] += 1
except IndexError:
continue
pt_cnt = pt_cnt / 2 * 255
pt_cnt[pt_cnt > 255] = 255
cv2.imwrite(os.path.join(feature_path, 'point.png'), pt_cnt)
def generate_dir_dist_data(traj_dir, grid_idx, feature_path):
MIN_DISTANCE_IN_METER = 5
MAX_DISTANCE_IN_METER = 300
dir_data = np.zeros((grid_idx.row_num, grid_idx.col_num, 8), dtype=np.uint8)
for filename in tqdm(os.listdir(traj_dir)):
if not filename.endswith('.txt'):
continue
traj_list = parse_traj_file(os.path.join(traj_dir, filename))
for traj in traj_list:
for i in range(len(traj.pt_list) - 1):
cur_pt, next_pt = traj.pt_list[i], traj.pt_list[i + 1]
if MIN_DISTANCE_IN_METER < distance(cur_pt, next_pt) < MAX_DISTANCE_IN_METER:
try:
row_idx, col_idx = grid_idx.get_matrix_idx(cur_pt.lat, cur_pt.lng)
direction = int(((bearing(cur_pt, next_pt) + 22.5) % 360) // 45)
dir_data[row_idx, col_idx, direction] += 1
except IndexError:
continue
np.save(os.path.join(feature_path, 'direction.npy'), dir_data)